Mudassar Abbas

Architected and developed a Gen AI chatbot for UDC-Qatar, reducing live agent workload and enabling 24/7 customer service. Implemented and productized high-value Gen AI use cases, including automated lead generation and service request handling, directly contributing to improved operational efficiency and sales pipeline velocity.

Developed NLP pipelines for large volumes of unstructured customer feedback, transforming raw text for analysis. Built and fine-tuned deep learning-based sales forecasting models (NBeats, DeepAR, Transformers) for predicting device sales, improving forecast accuracy.

Designed and deployed a computer vision pipeline (MTCNN, ResNET) for an inventory management system, automating asset logging and reducing manual tracking errors. Developed and implemented a deep learning solution for person tracking and demographic classification from CCTV footage to enhance retail analytics.

Co-developed and implemented the novel Virtual Adversarial Training (VAT) framework, a seminal contribution to semi-supervised deep learning. Published influential research in top-tier, peer-reviewed AI/ML conferences and journals.

Analyzed multi-source datasets (WHO, FAO) using Self-Organizing Maps (SOMs) to uncover key correlations between nutrition, lifestyle, and global health outcomes. Designed and implemented a multiple target tracking system for complex dynamic environments using state-space modeling frameworks.

Ensured the operational uptime and performance of business-critical subscriber services within a large-scale enterprise environment. Managed and monitored core platforms including Siebel CRM, TIBCO middleware, and Oracle databases, resolving issues to maintain service continuity for millions of subscribers.

Co-developed and implemented the novel Virtual Adversarial Training (VAT) framework, a seminal contribution to semi-supervised deep learning. Published influential research in top-tier, peer-reviewed AI/ML conferences and journals (e.g., Journal of Uncertainty Quantification, ICLR). Applied Bayesian Optimization and advanced statistical methods to complex modeling problems in climate science and n

Analyzed multi-source datasets (WHO, FAO) using Self-Organizing Maps (SOMs) to uncover key correlations between nutrition, lifestyle, and global health outcomes. Designed and implemented a multiple target tracking system for complex dynamic environments using state-space modeling frameworks. Developed statistical models to estimate latent parameters from real-world GPS sensor data, applying techni

Studied Computer & Information Sciences.